INDEX

Explanations

the name "Ett" with varying activation values

oai_token-act-pair · gpt-3.5-turbo

instances of the character string "tt" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dispers

-0.68

 sweeping

-0.67

 disperse

-0.65

 infiltrated

-0.62

 responsible

-0.61

 disbanded

-0.59

 solvent

-0.59

 conference

-0.58

 occupy

-0.57

 circulating

-0.57

POSITIVE LOGITS

tt

4.43

tta

1.91

tto

1.90

ttle

1.89

tti

1.74

TT

1.58

ttes

1.56

tty

1.54

tten

1.43

ts

1.38

Activations Density 0.011%

the name "Ett" with varying activation values

instances of the character string "tt" in various contexts

No Comments

No Known Activations